Care to elaborate? It's not obvious to me why/how that would happen. In fact, my experience so far tells me that code with tons of "why" comments (and possibly some "what"s as well) makes LLMs less likely to break stuff.